ng2-cells are not the cell of origin for murine...

ORIGINAL ARTICLE

NG2-cells are not the cell of origin for murine neurofibromatosis-1(Nf1) optic gliomaAC Solga, SM Gianino and DH Gutmann

Low-grade glial neoplasms (astrocytomas) represent one of the most common brain tumors in the pediatric population.These tumors frequently form in the optic pathway (optic pathway gliomas, OPGs), especially in children with theneurofibromatosis type 1 (NF1)-inherited tumor predisposition syndrome. To model these tumors in mice, we have previouslydeveloped several Nf1 genetically-engineered mouse strains that form optic gliomas. However, there are three distinct macroglialcell populations in the optic nerve (astrocytes, NG2þ (nerve/glial antigen 2) cells and oligodendrocytes). The presence of NG2þcells in the optic nerve raises the intriguing possibility that these cells could be the tumor-initiating cells, as has been suggested foradult glioma. In this report, we used a combination of complementary in vitro and novel genetically-engineered mouse strainsin vivo to determine whether NG2þ cells could give rise to Nf1 optic glioma. First, we show that Nf1 inactivation results in acell-autonomous increase in glial fibrillary acidic proteinþ (GFAPþ ), but not in NG2þ , cell proliferation in vitro. Second, similarto the GFAP-Cre transgenic strain that drives Nf1 optic gliomagenesis, NG2-expressing cells also give rise to all three macrogliallineages in vivo. Third, in contrast to the GFAP-Cre strain, Nf1 gene inactivation in NG2þ cells is not sufficient for opticgliomagenesis in vivo. Collectively, these data demonstrate that NG2þ cells are not the cell of origin for mouse optic glioma,and support a model in which gliomagenesis requires Nf1 loss in specific neuroglial progenitors during embryogenesis.

Oncogene advance online publication, 14 January 2013; doi:10.1038/onc.2012.580

Keywords: genetically-engineered mice; NG2þ ; optic glioma; neurofibromatosis type 1

INTRODUCTIONBrain tumors are the most common solid tumor in the pediatricpopulation, with low-grade glial neoplasms (pilocytic astrocytoma)representing the major histologic subtype in children between theages of five and fifteen (Central Brain Tumor Registry of the UnitedStates (CBTRUS) 2012; www.cbtrus.org). In contrast to adults, low-grade gliomas tend to arise in distinct brain regions, including thecerebellum, brainstem and optic pathway. This unique geographicpredilection is nicely illustrated by the pattern of glioma formationin individuals with the neurofibromatosis type 1 (NF1)-inheritedtumor syndrome. Fifteen to twenty percent of children with NF1will develop a low-grade glioma, the majority of which arise in theoptic pathway (optic pathway glioma, OPG).1,2 These OPGs canarise anywhere along the optic pathway, from the retro-orbitaloptic nerve to the post-chiasmatic optic tract.3,4 Most OPGs areslow-growing tumors composed of glial fibrillary acidic protein(GFAP)-immunoreactive cells with low proliferative indices (o1%).Whereas death from these tumors is rare, there is significantassociated morbidity, including hypothalamic dysfunction andvisual loss. In this regard, nearly half of children with NF1-associated OPGs will have visual impairment at initialpresentation.5

To gain insights into the molecular and cellular pathogenesis ofNF1-associated optic glioma, we and others have developed Nf1genetically-engineered mouse strains. Based on their glialhistology, Nf1 loss in GFAP-immunoreactive cells has beenmodeled using GFAP-Cre mouse lines. In these experiments,Nf1þ /� mice with GFAP-Cre-mediated Nf1 inactivation developoptic glioma.6,7 Careful analysis of the GFAP-Cre strains used in

these studies has revealed that Cre expression first occurs inGFAPþ neuroglial progenitor cells either at E11.57 or E14.5,8

rather than in differentiated astrocytes. These findings support amodel in which Nf1 loss must occur in specific neuroglialprogenitors during embryonic development in order forgliomagenesis to ensue.

In the optic nerve and relevant ventricular (germinal) zones,there are two types of potential neuroglial progenitors, GFAPþ 9,10

and nerve/glial antigen 2þ (NG2þ ) cells.11 This latter populationhas been shown to represent a potential cell of origin for ratmalignant gliomas,12,13 suggesting that NG2þ progenitors mayrepresent the initiating cell for Nf1 optic glioma. To determinewhether NG2þ neuroglial progenitors could serve as the cell oforigin for Nf1 genetically-engineered mouse optic glioma, weemployed a combination of in vitro and in vivo strategies. In thisreport, we demonstrate that Nf1 loss in NG2þ cells in vitro doesnot increase glial cell proliferation and that Nf1 loss in NG2þprogenitor cells in vivo is insufficient for optic gliomagenesis.Together, these data exclude NG2þ cells as the likely cell of originfor NF1-associated optic glioma and establish a model ofgliomagenesis in which Nf1 loss occurs in specific progenitorsduring embryonic development.

RESULTSThe mouse optic nerve is composed of three distinct types ofmacroglial cellsIn order to better characterize the macroglial compartment thatcontributes to optic gliomagenesis, we performed immunostaining

Department of Neurology, Washington University School of Medicine, St Louis, MO, USA. Correspondence: Dr DH Gutmann, Department of Neurology, Washington UniversitySchool of Medicine, Box 8111, 660 South Euclid Avenue, St Louis, MO 63110, USA.E-mail: [email protected] 16 May 2012; revised 26 October 2012; accepted 26 October 2012

Oncogene (2013), 1–11& 2013 Macmillan Publishers Limited All rights reserved 0950-9232/13

www.nature.com/onc

http://dx.doi.org/10.1038/onc.2012.580

www.cbtrus.org

mailto:[email protected]

http://www.nature.com/onc

with antibodies that recognize GFAP (astrocytes), nerve/glialantigen 2 (NG2 cells) and adenomatous polyposis coli (APC;oligodendrocytes). We found that the majority of macroglia in bothwild type (WT) and optic glioma-bearing (Nf1flox/mut; GFAP-Cre; OPGmice) mouse optic nerves are APCþ oligodendrocytes at both 3weeks and 3 months of age. In contrast, GFAPþ and NG2þ cells

compromise a smaller percentage of optic nerve macroglial cells(Figure 1a and Supplementary Figure 1). Importantly, upon Nf1loss, we observed a two-fold increase in the number of GFAPþastrocytes in the optic nerves of OPG mice relative to their WTcounterparts. The number of NG2þ cells and oligodendrocytesdid not change after Nf1 inactivation (Figure 1a).

WT

OP

G

GFAP NG2 APC

GFAP+0

10

20

30

40

% D

AP

I

0

10

20

30

40

% D

AP

I

Antibody cell type

Olig2 26

PDGFRα 54

APC

PDGFRβ 11

SMAα 68

oligodendrogliallineage

pericytes

WT

OP

G

GFAP, NG2 APC, NG2 GFAP, APC

NG2+ APC+

GFAP+ NG2+ APC+

% of NG2+ cells

4

Figure 1. Optic nerve astroglial cell populations in WT and Nf1 OPG mice. (a) Nf1flox/mut; GFAP-Cre (OPG) mice develop optic nervegliomas. Representative images of the optic nerves from 3-month old WT and OPG mice are shown. The arrow denotes an enlarged opticnerve and chiasm in one representative OPG mouse, not seen in WT mice. Histological comparison of cell type-specific markers demonstratesthat B30% of the cells are APCþ , 11% of the cells are NG2þ and 7% of the cells are GFAPþ. Increased numbers of GFAPþ astroglialcells are found in the optic nerves of OPG mice compared to WT mice. Each error bar represents mean±s.e.m. (b) Double-labeling of3-month-old WT and OPG nerves shows that these three glial cell populations are distinct. APC/NG2 double-positive cells account for fewerthan 5% of the cells in the optic nerve. Representative images are shown. Scale bar 50 mm. DAPI (blue) was used as a counterstain toidentify all cells in the sections. (c) NG2 double-labeling experiments revealed that 68 and 11% of the NG2 cells are SMAaþ or PDGFRbþ ,respectively (pericyte markers), whereas 26 and 54% of the NG2þ cells are Olig2þ and PDGFRaþ , respectively (oligodendroglial lineagemarkers).

NG2-glia and optic gliomaAC Solga et al

2

Oncogene (2013), 1 – 11 & 2013 Macmillan Publishers Limited

To establish that these macroglia represent distinct cell types,we performed double-labeling immunohistochemistry. In theseexperiments, there were no GFAPþ /APCþ or GFAPþ /NG2þcells, and fewer than 5% of the APCþ cells were NG2-immunopositive (Figures 1b and c). Next, we demonstrated thatnearly 100% of GFAPþ cells also co-expressed aldehydedehydrogenase 1 family, member L1 (ALDH1L1), previouslyreported as a marker of adult rat astrocytes.14 Similar to theGFAP immunostaining, we did not detect ALDH1L1þ /NG2þdouble-positive cells (Supplementary Figure 2).

To better characterize the NG2þ cell population in the opticnerve, we performed additional experiments based on previousstudies on NG2þ cells from other brain regions, suggesting thatNG2þ cells can be either pericytes15 or oligodendrocyteprecursors (OPCs).16 First, we demonstrated that the majority ofNG2þ cells (68%) in the normal optic nerve co-label with smoothmuscle actin a (SMAa; pericyte marker), whereas only 11% co-label with platelet-derived growth factor receptor b (PDGFRb;pericyte progenitor marker). However, 26% of the NG2þ cells inthe normal optic nerve co-labeled with Olig2, while 54% of theNG2þ cells were also PDGFRaþ . Few of the NG2þ cells wereAPC immunopositive (Figure 1c and Supplementary Figure 3A).These findings highlight the issues related to confidently assign-ing a cellular identity to NG2þ cells using available antibodies.

Second, to determine the identity of the NG2þ cells inNf1flox/mut; GFAP-Cre mouse optic gliomas, we performed NG2 andSMAa double labeling (Supplementary Figure 3B). Nearly 100% ofthe NG2þ cells were SMAaþ . A similar SMAa-staining patternwas also observed in a representative human NF1-associated OPG(Supplementary Figure 3C); however, NG2/SMAa double labelingcould not be performed in human tissue owing to technicallimitations of available NG2 antibodies for human tissue.

Third, as pericytes are often clustered around blood vessels, weperformed NG2/endoglin double labeling to determine whetherthere was preferential clustering of NG2þ cells near endothelialcells in Nf1flox/mut; GFAP-Cre mouse optic gliomas (SupplementaryFigure 3D). In these experiments, we observed a uniformdistribution of NG2þ cells in the optic nerve with no preferentialassociation near endoglinþ blood vessels.

To define the populations of macroglial cells that increase theirproliferation following Nf1 gene inactivation in OPG mice, weperformed Ki67 double-labeling experiments. Following GFAP-Cre-mediated Nf1 loss, we observed a three-fold increase in cellproliferation (Ki67-labeling index) relative to WT mice (Figure 2a).To determine which macroglial cells undergo GFAP-Cre-mediatedNf1 inactivation, we performed lineage tracing experiments usingthe Rosa-GREEN reporter strain. We found that all three macroglialcells (APCþ oligodendrocytes, NG2þ glia, and GFAPþ astro-cytes) exhibited Cre-mediated enhanced green fluorescent protein(EGFP) expression (Supplementary Figure 4), and therefore, couldrepresent the Nf1-deficient preneoplastic/neoplastic cell popula-tion relevant to optic gliomagenesis in mice.

Based on these findings, to determine which macroglial cell inthe murine Nf1 optic gliomas had increased proliferation followingNf1 gene inactivation, double immunofluorescence studies wereperformed. Consistent with the designation of these tumors asastrocytomas, there were no proliferating APCþ cells (oligoden-drocytes). However, both NG2þ and GFAPþ cells exhibitedincreased proliferation in the optic nerves of OPG mice relative tolittermate controls at 3 months of age (Figure 2b). Similar resultswere obtained using ALDH1L1 as an additional astrocyte marker(Supplementary Figure 5). The small number of proliferating cellsin prechiasmatic optic nerve and chiasm of WT mice (o2 Ki67þcells per specimen) precluded a meaningful comparison.Together, these findings demonstrate that only NG2þ andGFAPþ , but not APCþ , macroglial cells hyperproliferate follow-ing Nf1 loss in OPG mice in vivo.

Only GFAPþ optic nerve astrocytes exhibit increased proliferationin response to Nf1 inactivation in vitroPrevious studies from our laboratory revealed that astrocytecultures from the optic nerve are cellularly- and molecularly-distinct macroglial cell populations.11 Whereas primary postnatal(PN) day 1 astrocyte cultures from the brainstem, neocortex, andcerebellum are composed of 498% GFAPþ cells with fewer than5% NG2þ cells, optic nerve astrocyte cultures under the identicalin vitro conditions are composed of B30% GFAPþ and B70%NG2þ cells (Figure 3a). Similar to macroglial cells in the intactmouse optic nerve, there were no glial cells expressing both GFAPand NG2. Under these culture conditions, no O4þ oligodendro-cytes were generated.

To determine which glial cell population had the capacityto hyperproliferate in response to Nf1 gene inactivation, PN1Nf1flox/flox optic nerve astroglial cultures were generated (passage0) and infected with either Ad5-Cre or Ad5-LacZ virus to generateNf1-deficient (Nf1� /� ) and WT astroglial cells (passage 2),respectively. Following Cre-mediated Nf1 inactivation, neurofibro-min was undetectable in these cultures (Figure 3b) concomitantwith an overall 1.5-fold increase in proliferation (Figure 3c).Whereas Nf1 loss resulted in a B1.5-fold increase in proliferationof GFAPþ cells, no increased proliferation was observed in theNG2þ cell population relative to Ad5–LacZ–infected cultures(Figure 3d). It is possible that differences in Nf1 gene expressionunderlie the failure of NG2þ cells to increase their proliferation

GFAP+ NG2+ APC+0

10

20

30

40

% K

i-67

+ ce

lls

OPG

WT

WT OPG0.0

0.5

1.0

1.5

*

Ki6

7 la

bel

ing

ind

ex

Figure 2. Neurofibromin loss results in increased NG2 and GFAP cellproliferation in vivo. (a) Proliferation was measured by Ki-67 labeling.Nf1 inactivation results in a three-fold increase in the percent of Ki-67 positive cells (labeling index) in the optic nerves of OPG micecompared to WT mice at 3 months of age. Representative imagesare shown. Arrows denote representative Ki-67-immunopositivecells. Scale bar 100 mm. (b) The percentages of GFAP/Ki-67, NG2/Ki-67 and APC/Ki-67 cells in the optic nerves of OPG mice at 3 monthsof age are shown. The images above each bar in the graph depictrepresentative GFAP/Ki-67 double-positive, NG2/Ki-67 double-posi-tive, or APC-positive/Ki-67-negative cells, respectively. Each error barrepresents mean±s.e.m. Asterisks denote statistically significantdifferences (*) P¼ 0.0482.


3

& 2013 Macmillan Publishers Limited Oncogene (2013), 1 – 11

following Nf1 loss. To determine whether NG2þ and GFAPþ cellshave different relative levels of Nf1 mRNA expression, weemployed Nf1 RNA fluorescence in situ hybridization (FISH). Inthese experiments, GFAPþ cells and NG2þ cells had comparablelevels of Nf1 mRNA expression (Figure 3e). Collectively, theseresults demonstrate that GFAPþ astrocytes are the macroglialpopulation with the greatest ability to hyperproliferate followingneurofibromin loss in vitro.

NG2þ cells give rise to oligodendrocytes and astrocytes in vivoTo further analyze the NG2þ cell population in mouse optic nerveglial cell cultures, we performed several additional experiments.First, we used double-labeling immunofluorescence to demon-strate that nearly all NG2þ cells co-label with PDGFRb, while 64%of the NG2þ cells are SMAaþ (pericyte markers). Moreover, 40and 67% of the NG2þ cells are double-positive for A2B5 and

PDGFRa, respectively (OPC markers) (Figure 4a). Second, todetermine the capacity of these NG2þ cells to differentiate intodistinct glial cell populations, we employed immunopanning toobtain highly purified NG2þ cells from the optic nerve. Thispurified NG2þ cell population remained 100% NG2þ after oneday in culture, without any contaminating GFAPþ astrocytes orO4þ oligodendrocytes, but did not generate either O4þ orGFAPþ cells under either serum or serum-free conditions after 3and 7 days in culture (Figure 4b). These findings again underscorethe problems associated with confidently distinguishing NG2þcell populations using cell type-specific antibodies, and demon-strate that NG2þ optic nerve cells are unlikely to be glialprogenitors in vitro.

Third, to determine whether NG2þ cells could give rise to thethree major macroglial populations in the optic nerve as has beenreported for the forebrain,17 we leveraged a mouse strain in whichCre recombinase is expressed from the endogenous NG2

α-tubulin

WT Nf1-/-

neurofibromin

GFAP, NG2 NG2+ GFAP+0

20

40

60

80

% D

AP

I

0

5

10

15 *

% B

rdU

+ c

ells

WT Nf1-/-

GFAP+ NG2+0.000

0.002

0.004

0.006N

f1 m

RN

A /

�m2

GFAP, Nf1 mRNA NG2, Nf1 mRNA

WT

Nf1-/-

N. S.

GFAP+ NG2+0

10

20

30

40

50

60

% B

rdU

+ c

ells

**

WT Nf1-/-

GF

AP

NG

2

Figure 3. Neurofibromin loss results in increased optic nerve GFAPþ cell proliferation in vitro. (a) Double-labeling immunocytochemistryreveals that 30% of the cells in optic nerve glial cell cultures are GFAPþ, whereas 70% of the cells are NG2þ . There are no GFAP/NG2 double-positive cells. Following Nf1 inactivation by Ad5-Cre expression, loss of neurofibromin expression was observed (b) coincident with anincrease in proliferating (BrdUþ ) cells (c). a-tubulin is used as an internal control for protein loading. (d) Following Nf1 inactivation by Ad5-Creexpression, only the GFAPþ population of cells exhibited increased proliferation. Representative BrdUþ proliferating cells (green) andDAPIþ nuclei (blue) are shown. The percentage of BrdUþ /NG2þ and BrdUþ /GFAPþ cells are represented as the mean±s.e.m. Scale bar,100 mm. N. S.¼not significant, (**) P¼ 0.0055, (*) P¼ 0.0129. (e) FISH revealed comparable levels of Nf1 mRNA expression (green particles;inset) in GFAPþ and NG2þ cells. The number of signals is the average of 10 counted cells per area and is displayed in the graph. Nuclei werecounterstained with DAPI (blue). Representative images are shown. Scale bar, 50 mm.


4


promoter.17 NG2-Cre mice were crossed with Rosa-GREENreporter mice, and at 3 weeks of age, we detected EGFPexpression in NG2þ , APCþ and ALDH1L1þ cells (Figure 5),similar to that observed with the GFAP-Cre strain used to generateNf1 optic gliomas (Supplementary Figure 4). Similar to the in vitrofindings, we also found SMAaþ cells with NG2-mediated EGFPexpression in the optic nerve (Supplementary Figure 6A). SinceA2B5 cannot be employed for immunohistochemistry, we derivedoptic nerve cultures from Rosa-GREEN�NG2-Cre mice to demon-strate that NG2þ cells can give rise to A2B5þ cells in the opticnerve (Supplementary Figure 6B). Collectively, these resultsdemonstrate that NG2þ cells give rise to all macroglial celllineages in vivo.

Nf1flox/mut; NG2-Cre mice do not develop optic gliomaTo determine whether Nf1 loss in NG2þ progenitor cells issufficient for optic gliomagenesis in mice, we generated Nf1þ /�mice with NG2-specific Nf1 inactivation. This genetic configurationis similar to the Nf1 optic glioma mice in which Nf1þ /- miceharbor neurofibromin loss in GFAPþ cells.6 Cohorts of tenNf1flox/mut; NG2-Cre mice and ten control littermates (NG2-Cre andNf1flox/WT; NG2-Cre mice; n¼ 10 each) were collected. These micewere healthy and bred successfully.

Analysis of the optic nerves from Nf1flox/mut; NG2-Cre micerevealed no evidence of optic glioma formation at 3 months ofage. Specifically, we observed no gross morphological or histologicaldifferences and only a slight increase in optic nerve volume.

There was no change in mitotic indices, microglia numbers,GFAPþ astrocyte numbers or NG2þ glial cell numbers (Figure 6).Moreover, no optic gliomas were observed using the abovecriteria in a cohort of Nf1flox/mut; NG2-Cre mice at 6 months of age(data not shown; n¼ 7).

The failure to generate optic gliomas in Nf1flox/mut; NG2-Cre micecould reflect a number of possibilities, including the efficiency ofCre-mediated recombination as well as the timing and specific celltype in which Nf1 loss occurs. To demonstrate Nf1 inactivation inthe optic nerve, we first employed recombination PCR on wholetissue (Figure 7a). Both Nf1flox/mut; NG2-Cre and Nf1flox/mut; GFAP-Cre mouse optic nerves exhibited Nf1 gene inactivation. Second,we employed the Rosa-GREEN reporter mice to show that bothNG2-Cre and GFAP-Cre mouse optic nerves had equivalentnumbers of EGFPþ cells, indicative of Cre-mediated excision(Figure 7b). Third, we used RNA in situ hybridization to measureNf1 mRNA expression Nf1flox/mut; NG2-Cre and Nf1flox/mut; GFAP-Cremouse optic nerves. There was a similar reduction in total Nf1mRNA expression in these mice relative to their WT counterparts(42% decrease in Nf1flox/mut; NG2-Cre and 67% decrease inNf1flox/mut; GFAP-Cre mouse optic nerves) (Figure 7c). As we couldnot confidently localize the Nf1 mRNA FISH signal to a specific celltype by double-labeling immunofluorescence, we performedGFAP/neurofibromin and NG2/neurofibromin double labeling. Inoptic nerve specimens from both Nf1flox/mut; NG2-Cre andNf1flox/mut; GFAP-Cre mice, there was a 30–37% reduction in thepercent of GFAP/neurofibromin double-positive cells relative toWT mice, with no change in the percent of NG2/neurofibromin

day 1 day 3 day 7

optic nerve

tissue dissociation

NG2-immunopanning

days1 3 7

NG2, PDGFRα

Antibody % of NG2+ cells cell typeOlig2 0A2B5 40

PDGFRα 67O4 0

PDGFRβ 100SMAα 64

oligodendrogliallineage

pericytes

PDGFRβ, GFAP

NG2, A2B5

NG2, SMAα

Figure 4. Characterization of NG2þ optic nerve cells in vitro. (a) 64% of the NG2 cells are double-positive for SMAa and almost all arePDGFRbþ (pericyte markers). None of the GFAPþ cells are SMAaþ or PDGFRbþ . In contrast, 40 and 67% of the NG2þ cells are A2B5þ orPDGFRaþ , respectively (oligodendroglial lineage markers). (b) Following NG2 antibody immunopanning and replating, PN1 optic nerveNG2þ cells grown for 1, 3 and 7 days do not differentiate into either GFAPþ or O4þ cells in vitro. Nuclei were counterstained with DAPI(blue). Representative images are shown. Scale bar, 100 mm.


5


double-positive cells (Figure 7d and Supplementary Figure 7).These data exclude differences in Nf1 gene expression orinactivation as the likely reason for the failure of Nf1flox/mut;NG2-Cre mice to develop optic glioma.

Next, to determine when and where the NG2-Cre transgene isfirst expressed relative to the GFAP-Cre transgene used togenerate Nf1 optic glioma mice, NG2-Cre and GFAP-Cre micewere crossed to Rosa-GREEN reporter mice and their brainsexamined at various developmental times. In these experiments,Cre activity was first detected in the third (TVZ) and lateralventricular zones (lv-SVZ) at embryonic day 14.5 (Figure 8a),similar to the GFAP-Cre mouse strain (Figure 8b), excludingdifferences in the timing of Nf1 loss.

Finally, we sought to determine where the NG2-Cre transgenewas expressed in the lateral and third ventricle germinal zonesin vivo. Whereas both NG2-Cre and GFAP-Cre mice generateEGFPþ /nestinþ cells, the EGFPþ /nestinþ cells in the NG2-Crestrain were not preferentially localized to the periventricularareas around the lv-SVZ and TVZ (Figure 8a). Moreover, therewere EGFPþ /PDGFRaþ and EGFPþ /PDGFRbþ cells similarlydistributed throughout the TVZ (Supplementary Figure 6C). Incontrast, EGFPþ /nestinþ cells in the GFAP-Cre mouse werepreferentially localized to the periventricular zones alongthe wall of the third ventricle (Figure 8b), consistent with theirdesignation as potential neural stem cells relevant to opticglioma.18 Collectively, these findings argue that NG2þ progenitorcells are not identical to GFAP-Cre optic glioma progenitorswith respect to their ventricular location, and support amodel in which Nf1 inactivation likely occurs in specificprogenitors (GFAP-Cre-expressing stem cells) to result in opticgliomagenesis.

DISCUSSIONNumerous studies have shown that oligodendrocytes, astrocytesand NG2þ cells comprise the three types of macroglial cells in themammalian central nervous system. In most brain regions,astrocytes are the major macroglial cell population, whereasNG2-glia comprise 8–9% of the total cell population in whitematter regions, and 2–3% of the total cells in gray matterregions.19 Relevant to low-grade gliomas in children, similarmacroglial cell populations exist in the optic nerve. The opticnerve is a heavily myelinated structure composed of axonscarrying visual information from the retina to the lateral geniculatebodies. For this reason, oligodendrocytes constitute the majorityof the macroglial cells in the optic nerve.

However, optic gliomas are not oligodendroglial neoplasms(oligodendrogliomas), and are instead composed of GFAP-immunoreactive cells. Consistent with observations in humantumors, we found that the majority of the proliferating cells in theNf1 mouse optic glioma were GFAPþ or NG2þ cells, rather thanAPCþ cells (oligodendrocytes). In addition, we found that glialcell cultures from the normal optic nerve were composed of twodistinct non-oligodendrocyte cell types that express either NG2 orGFAP, but not both markers,11 analogous to studies performed oncells from other regions of the central nervous system. For thisreason, we focused on the relative contributions of GFAPþ andNG2þ cells to Nf1 murine optic gliomagenesis.

Early landmark studies examining rat optic nerve glia revealedtwo types of astrocytes (type-1 and type-2).20 In contrast to type-1astrocytes, type-2 astrocytes are defined by A2B5 expression20,21

and are thought to arise from A2B5þ , but GFAP- andgalactocerebrosidase-negative, oligodendrocyte-type-2 astrocyteprecursor cells.20–22 Unfortunately, A2B5þ cells cannot be reliably

Cell type-specific marker EGFP Merge

NG

2A

PC

ALD

H1L

1

Figure 5. NG2-Cre cells give rise to NG2þ , ALDH1L1þ and APCþ glial cell populations in vivo. Optic nerve sections from Rosa-GREEN�NG2-Cre mice reveal EGFPþ /NG2þ cells, EGFPþ /APCþ (oligodendrocytes) and EGFPþ /ALDH1L1þ (astrocytes). Scale bar, 50 mm.Representative images are shown with insets of immunopositive cells.


6


detected in vivo owing to antibody technical limitations. Based onthe fact that some rat optic nerve NG2þ cells co-express A2B5and are capable of generating GFAPþ astrocytes23 and A2B5þcells24 in vitro, it has been postulated that a population ofNG2þ cells may serve as O2A progenitors. The progenitorproperties of NG2þ cells is reinforced by studies in the spinalcord in which some NG2þ cells express nestin or A2B5, and cangive rise to oligodendrocytes and astrocytes.25,26 Additionally,embryonic-mouse forebrain neurospheres contain a smallpopulation of NG2/A2B5 double-positive cells.27 However, using

NG2-Cre mice, none of the GFAPþ cells originated from NG2þcells in white matter regions in vivo.17

In support of the role for NG2þ cells as glial progenitors in theoptic nerve, we found that 40% of the NG2þ cells in the mouseoptic nerve express A2B5 in vitro, and that NG2þ cells give rise toastrocytes in vivo. Moreover, we found that 26 and 54% of theNG2þ cells in the optic nerve co-express either Olig2 or PDGFRa,respectively, markers of glial progenitor cells. The ability of NG2þcells to function as progenitors for astrocytes in some regions ofthe central nervous system (optic nerve, spinal cord), but not in

NG2-Cre

0

Ki6

7Ib

a1G

FA

P

0

1

2

3

4

5

NG2-Crevo

lum

eNf1flox/wt; NG2-Cre Nf1flox/wt;

NG2-CreNf1flox/mut; NG2-Cre Nf1flox/mut;

NG2-Cre

0.00

0.02

0.04

0.06

0.08

0.10

NG

2

0

10

20

% N

G2+

cel

ls

**

*

volu

me

(mm

3 )K

i67

lab

elin

g in

dex

1.0

0.8

0.6

0.4

0.2

0.0

%Ib

a1+c

ells

%G

FAP

+cel

ls30

20

10

Figure 6. Nf1flox/mut; NG2-Cre mice do not develop optic glioma. Representative images of the optic nerves from all three genotypes areshown. Nf1flox/mut; NG2-Cre mice optic nerve volumes are slightly increased relative to control (NG2-Cre and Nf1flox/WT; NG2-Cre) mice. However,no increases in the Ki67-labeling index, percent of Iba1þ microglia, or percent of GFAPþ or NG2þ cells were found in Nf1flox/mut; NG2-Cremice relative to control mice. Scale bar, 100 mm, 50mm NG2 panel. Error bars represent mean±s.e.m. Asterisks denote statistically significantdifferences (*) P¼ 0.0119, (**) Po0.0052.


7


other locations (white matter regions), highlights the innateheterogeneity of this population of cells.

NG2 has also been shown to label pericytes, which function asvital integrators of the neurovascular unit and coordinate a varietyof homeostatic functions, including angiogenesis, blood–brain–barrier integrity, and clearance of toxic cellular byproducts.28 Todetermine whether the NG2þ cells in the optic nerve werepericytes, we performed double-labeling experiments using twodifferent pericyte markers (SMAa and PDGFRb). Interestingly,NG2þ cells in the optic nerve and brain co-label with both ofthese markers. Consistent with this finding, our previousmicroarray studies similarly revealed increased expression ofPDGFRb in glial cell cultures from the optic nerve compared toother brain-region primary astrocyte populations.11 In addition, wefound that some of the NG2þ cells in the optic nerve were inclose proximity to blood vessels, whereas others were uniformlydistributed within the parenchyma. The fact that NG2þ cellsexpress markers associated with pericytes as well as glialprecursors highlights the difficulties associated with the use ofcurrently available antibodies to distinguish between these NG2þcell populations.

Support for the notion that NG2þ cells are potential gliomaprogenitor cells derives from immunohistochemical experiments

demonstrating increased NG2 expression in invasive mouseglioma29 and in the actively proliferating population of humanhigh-grade glioma,30 correlating with increasing gliomamalignancy grade.31 Similarly, we found robust SMAaimmunolabeling in one representative NF1-associated opticglioma. Second, retroviral PDGF-induced rat gliomas arising inthe spinal cord30 and brainstem13 contain a highly proliferativepopulation of NG2þ progenitor cells. Third, overexpression ofNG2 in mouse glioblastoma multiforme increases glioma growth,whereas NG2 knockdown decreases glioma growth in vivo.33

Fourth, A2B5þ /CD133� cells freshly isolated from human high-grade gliomas form tumors when injected intoimmunocompromised nu/nu mice.34,35 Fifth, introducing p53/Nf1mutations directly into OPCs cause malignant gliomas in adultmice.36 These latter experiments were performed using themosaic analysis with double markers approach, and revealedthat high-grade gliomas arose in 8-month-old mice following theinactivation of both the Nf1 and p53 genes in lv-SVZ OPCs. Similarto elegant studies by the Parada laboratory using mice withconditional Nf1, p53 and Pten inactivation,37 SVZ progenitors serveas the putative cells of origin for malignant glioma in adult mice.In contrast to these adult malignant glioma studies, we haverecently shown that low-grade Nf1 mouse optic gliomas likely

R

GFAP-Cre xRosa-GREEN

NG2-Cre xRosa-GREEN

0

5

10

15

20

25

% E

GF

P+

cells

WT

0

20

40

60

80

GFAP+ NG2+

% n

euro

fib

rom

in+

cells

*WT

0

500

1000

1500

Nf1

mR

NA

mo

lecu

les

*

WT Nf1flox/mut;GFAP-Cre

Nf1flox/mut; GFAP-Cre

Nf1flox/mut; GFAP-Cre

Nf1flox/m

ut ; GFA

P-Cre

Nf1flox/mut;NG2-Cre

Nf1flox/mut; NG2-Cre

Nf1flox/mut; NG2-Cre

Nf1flox/m

ut ; NG2-

Cre

Figure 7. Nf1 inactivation is observed in Nf1flox/mut; NG2-Cre mice. (a) Recombination PCR demonstrates Cre-mediated Nf1 gene recombination(R) in both Nf1flox/mut; GFAP-Cre mice and Nf1flox/mut; NG2-Cre mice. (b) The efficiency of Cre-mediated recombination was determinedfollowing intercrossing of GFAP-Cre and NG2-Cre with Rosa-GREEN reporter mice. The percent of EGFPþ cells relative to the total number ofcells in Rosa-GREEN�GFAP-Cre mice (13%) is similar to that observed in Rosa-GREEN�NG2-Cre mice (20%). (c) FISH analysis reveals similarreductions in Nf1 mRNA (red) expression in the optic nerves of Nf1flox/mut; GFAP-Cre mice (67%) and Nf1flox/mut; NG2-Cre mice (42%) comparedto WT mice. Nuclei were counterstained with DAPI (blue). Representative images are shown. Scale bar, 50 mm. (d) Nf1 deletion in Nf1flox/mut;NG2-Cre mice was determined by GFAP/neurofibromin and NG2/neurofibromin double-labeling using Nf1flox/flox mice as controls. Similar toNf1flox/mut; GFAP-Cre mice with optic glioma, Nf1flox/mut; NG2-Cre mouse optic nerves exhibit a 30% reduction in the percentage of GFAP/neurofibromin double-positive cells. No change in the percent of NG2/neurofibromin double-positive was observed in Nf1flox/mut; NG2-Cremice or Nf1flox/mut; GFAP-Cre mouse optic nerves. Each error bar represents mean±s.e.m. Asterisks denote statistically significant differences(*) P¼ 0.044.


8


arise from third ventricle progenitors, rather than lv-SVZprogenitors, during embryogenesis.18 Consistent with thisunique progenitor origin, Nf1 loss in TVZ neural stem cellsresulted in increased proliferation and glial differentiation,whereas Nf1 loss in lv-SVZ neural stem cells did not.

To experimentally define the role of NG2þ cells in Nf1 opticgliomagenesis, we employed a complementary series of in vitroand in vivo approaches. We initially demonstrated that NG2þcells constitute a significant fraction of the non-oligodendrocyteglial cells in the normal optic nerve as well as in Nf1 optic glioma.In the optic nerve, NG2þ cells represent a distinct population ofglial cells in vivo and in vitro, lacking GFAP expression. Upon acuteNf1 gene inactivation in vitro, NG2þ cells did not exhibit anincrease in proliferation, whereas increased proliferation wasobserved in the GFAPþ glial cell population. Next, we showedthat NG2þ cells from NG2-Cre mice in which Cre recombinase isexpressed from the endogenous NG2 promoter can give rise toGFAPþ cells in the optic nerve in vivo. Similar to GFAP-Cre-drivenNf1 inactivation, we found that NG2-Cre-driven Nf1 inactivationresulted in reduced neurofibromin expression in GFAPþ cells inthe optic nerve; however, in striking contrast to Nf1flox/mut; GFAP-Cre mice, Nf1þ /� mice in which Nf1 loss occurs in NG2þ cells

(Nf1flox/mut; NG2-Cre mice) do not form optic gliomas at 3 or 6months of age in vivo. Further experiments are needed to clarifywhy the NG2þ cells in the Nf1flox/mut; NG2-Cre and Nf1flox/mut;GFAP-Cre mouse optic nerves retain Nf1 expression, despite beingderived from NG2-Cre- and GFAP-Cre-expressing progenitor cells.

The absence of optic gliomagenesis could reflect severaldifferences between the GFAP-Cre and NG2-Cre mice used forthese experiments. First, it is possible that NG2-Cre mice exhibitless efficient Cre-mediated Nf1 inactivation. This possibility wasexcluded using Rosa-GREEN reporter mice, in which the percen-tage of cells with Cre-mediated excision is actually slightly higherthan that observed in GFAP-Cre mice. Moreover, optic nerves fromboth Nf1flox/mut; GFAP-Cre and Nf1flox/mut; NG2-Cre mice exhibitedsimilar levels of Nf1 inactivation as assessed by recombination PCRas well as RNA FISH. Second, optic nerve astrocytes might not arisefrom NG2þ progenitors. However, we showed that both GFAP-Cre and NG2-Cre mice are capable of generating astrocytes in theoptic nerve. Third, it is possible that Cre-mediated Nf1 loss occursat the wrong time point for glioma formation. Recent studiesfrom our laboratory have shown that gliomas are only formed inNf1þ /� mice when Nf1 gene loss occurs during embryogen-esis.18 As NG2-Cre-mediated recombination begins at E14.5,

EB

ON

lv-SVZ TVZ Eyeball / ON

EB

ON

EG

FP

EG

FP

nest

in, E

GF

Pne

stin

, EG

FP

Figure 8. NG2-Cre is first expressed at E14.5 in cells in the third and lateral ventricles. (a) NG2-Cre- and (b) GFAP-Cre-mediated EGFP expressionis first detected by embryonic day E14.5 in the lv-SVZ (dotted line) and TVZ (dotted line) of (a) Rosa-GREEN�NG2-Cre and (b) Rosa-GREEN�GFAP-Cre mice, respectively. Nestin/EGFP double-positive cells in Rosa-GREEN�NG2-Cre mice are randomly scattered in the areasaround the lv-SVZ and TVZ, whereas nestin/EGFP double-positive cells in Rosa-GREEN�GFAP-Cre mice are preferentially localized to the TVZ,but not to the lv-SVZ, periventricular region. Representative images are shown with insets of immunopositive cells. Scale bar, 100 mm. EB¼eye ball, ON¼optic nerve.


9


similar to GFAP-Cre mice, it is unlikely that temporal differencesaccount for the lack of optic gliomas in Nf1flox/mut; NG2-Cre mice.

Fourth, it is possible that NG2þ cells are not located in aprogenitor zone relevant to optic gliomagenesis. Recent studiesfrom our laboratory have shown that Nf1 murine optic gliomaslikely arise from progenitor cells lining the third ventricle, ratherthan the lateral ventricle.18 This ventricular zone has previouslybeen shown to contain the progenitor cells that give rise to opticnerve OPCs.38,39 In addition, a recent gene expression study usinghuman optic-glioma specimen supported the third ventricle as thelikely germinal zone of origin for these tumors.40 We havepreviously shown in both human and mice that the third ventriclecontains nestinþ radial glia-like cells,41 analogous to type B-cells,important for gliomagenesis in adult malignant glioma.10,41–43 Inthis report, we demonstrate, using a Rosa-GREEN reporter mousestrain, that unlike the GFAP-Cre strain used to generate Nf1 murineoptic gliomas, the EGFPþ /nestinþ cells from NG2-Cre mice werenot preferentially localized to the periventricular zone of the thirdventricle.

This intriguing difference suggests that the failure of Nf1flox/mut;NG2-Cre mice to develop optic glioma may reflect the require-ment for a specific population of susceptible progenitor cells(potential tumor-initiating cells) within the third ventricle germinalzone to expand following Nf1 inactivation, and generate thesecommon low-grade brain tumors in children.

MATERIALS AND METHODSMiceNf1flox/flox (WT)44 and Nf1flox/mut; GFAP-Cre6 mice were generated aspreviously described. For cell-fate mapping experiments, NG2-Cremice (B6,FVB-Tg(Cspg4-cre)1Akik/J, Jackson Laboratory, Bar Harbor, ME,USA17) and GFAP-Cre mice8 were intercrossed with Rosa-GREEN (B6.Cg-Gt(ROSA)26Sortm(CAGZsGreen1)Hzeþ /J; Jackson Laboratory) reporter mice,45

respectively. To establish Nf1þ /� background mice with NG2-specific Nf1inactivation, Nf1þ /� mice46 were bred with Nf1flox/WT mice to produceNf1flox/mut mice, which were subsequently intercrossed with Nf1flox/flox;NG2-Cre mice to generate Nf1flox/mut; NG2-Cre mice. Nf1flox/WT; NG2-Cremice were generated by crossing Nf1flox/flox mice with NG2-Cre mice. Allmice were maintained on a C57Bl/6 background and used in accordancewith approved animal studies protocols at the Washington UniversitySchool of Medicine.

Primary astrocyte culturesPrimary astrocyte cultures were established from the optic nerves ofPN day 1–2 Nf1flox/flox pups.11 WT and Nf1-deficient (Nf1� /� ) cultureswere generated following infection with Adenovirus type 5 containingb-galactosidase (Ad5-LacZ) or Cre recombinase (Ad5-Cre) (University ofIowa Gene Transfer Vector Core, Iowa City, IA, USA), respectively.Neurofibromin loss was confirmed by western blot.

Cell proliferationApproximately 50 000 astroglial cells were plated in 24-well dishes, allowedto adhere, and maintained in astrocyte growth medium for 24 h beforeexposure to the thymidine analog BrdU (5 mM) for 4 h.

NG2 immunopanningCell selection was based on the ‘panning’ technique described by Stallcupand Beasley.23 Briefly, optic nerves were removed from decapitated PN1–2pups and incubated for 30 min at 37 1C in 2 ml HEPES-buffered DMEMcontaining 1mg/ml collagenase (Sigma, St Louis, MO, USA). Followingextensive washing, the dissociated optic nerve cells were subjected toimmunopanning on six-well dishes precoated with 10 mg/ml NG2monoclonal antibodies for 30–45 min (Santa Cruz Biotechnology, SantaCruz, CA, USA). Unbound cells were removed and cultured independentlyfor comparison with bound cells. Bound cells were subsequently detachedfrom the dishes by scraping. All cells were washed once and plated in cellculture plates. For the differentiation assay, cells were grown for 1, 3 or 7days in astrocyte growth medium.

ImmunocytochemistryAstrocytes were fixed in 4% paraformaldehyde and permeabilized with0.2% Triton X-100. Following overnight incubation with primary antibodies(Supplementary Table 1), visualization was accomplished after incubationwith either Alexa Fluor 488 or 568 IgG or IgM secondary antibodies(Invitrogen, Grand Island, NY, USA). Cells were counterstained with DAPI.To detect BrdU incorporation, cells were treated with 2 M HCl to denatureDNA followed by overnight incubation with monoclonal anti-BrdUantibodies (1:200).

For each independent culture, at least 5 distinct microscopic fields wereanalyzed on a Nikon Eclipse TE300 fluorescence inverted microscope(Nikon, Tokyo, Japan) equipped with an optical camera (Optronics, Goleta,CA, USA) and MetaMorph image analysis software (Molecular Devices,Dowingtown, PA, USA).

ImmunohistochemistryFor immunohistochemistry on paraffin sections, horseradish peroxidase-conjugated secondary antibodies (Vector Laboratories, Burlingame, CA,USA) were used in combination with Vectastain Elite ABC developmentand hematoxylin counterstaining. For immunofluorescence detection,appropriate Alexa-Flour tagged secondary antibodies (Invitrogen) wereused, followed by DAPI counterstaining.

RNA in situ hybridizationFISH was performed using the QuantiGene ViewRNA kit (Affymetrix Inc.,Frederick, MD, USA) according to the manufacturer’s instructions withminor modifications. The proteinase K treatment was omitted in order topreserve the integrity of the glial cells. The oligonucleotide probe wasdesigned commercially using the murine Nf1 sequence (accession numberNM_010897.2). Following FISH, cells were incubated in blocking buffer(10% goat serum in 0.3% PBS-Triton X-100) for 1 h, and subsequentlyprocessed for immunofluorescence using standard methods. Images wereobtained with a Nikon Eclipse TE300 fluorescence inverted microscope(Nikon) and analyzed using MetaMorph image analysis software (MolecularDevices, Dowingtown, PA, USA). Briefly, cell areas were measured bytracing the cell body, so that the cell body size would not affect themeasurements. The Nf1 mRNA punctae in the cell body were counted andmRNA molecules per area were calculated. For FISH in tissue, theconditions were optimized to include a 10 min boiling and a 10 minprotease treatment.

Western immunoblottingCells were lysed in 1% NP-40 lysis buffer, supplemented with protease andphosphatase inhibitors, and protein concentrations determined using theBCA protein assay (Pierce, Thermo Scientific, Rockford, IL, USA). FollowingSDS-PAGE separation and western blotting, neurofibromin expression wasdetected using rabbit anti-NF1GRP-D antibodies (1:200 dilution; Santa CruzBiotechnology). Detection was accomplished by enhanced chemilumines-cence using the ChemiDoc-It Imaging System (UVP, Upland, CA, USA). a-tubulin (Sigma) served as an internal control for protein loading anddensitometric normalization.

Optic nerve measurementsOptic nerves with an intact chiasm were microdissected, photographed,and optic nerve diameters measured at the chiasm at B200 and B400and B600m anterior to the chiasm to generate volumes as previouslyreported.47

Statistical analysisAll in vitro experiments were repeated at least three times with similarresults. Statistical analysis was performed using GraphPad Prism 4.0software (GraphPad, La Jolla, CA, USA). Data are presented as mean valueswith s.e.m. Statistical significance was assessed by using Student’stwo-tailed t-test. Grubbs outlier test was used to determine statisticaloutliers. Statistical significance was set at Po0.05.

CONFLICT OF INTERESTThe authors declare no conflict of interest.


10


ACKNOWLEDGEMENTSWe appreciate the excellent technical assistance of Angela Petti and BelindaMcMahan in the Ophthalmology Core Facility for preparing the optic nerve sections.This work was funded by a grant from the National Cancer Institute (U01-CA141549to DHG), whereas the Ophthalmology Core Facility is funded by a grant from theNational Eye Institute (EY02687).

REFERENCES1 Blazo MA, Lewis RA, Chintagumpala MM, Frazier M, McCluggage C, Plon SE.

Outcomes of systematic screening for optic pathway tumors in children withneurofibromatosis type 1. Am J Med Genet A 2004; 127A: 224–229.

2 Lund AM, Skovby F. Optic gliomas in children with neurofibromatosis type 1. Eur JPediatr 1991; 150: 835–838.

3 Listernick R, Darling C, Greenwald M, Strauss L, Charrow J. Optic pathway tumorsin children: the effect of neurofibromatosis type 1 on clinical manifestations andnatural history. J Pediatr 1995; 127: 718–722.

4 Czyzyk E, Jozwiak S, Roszkowski M, Schwartz RA. Optic pathway gliomas in chil-dren with and without neurofibromatosis 1. J Child Neurol 2003; 18: 471–478.

5 Listernick R, Louis DN, Packer RJ, Gutmann DH. Optic pathway gliomas in childrenwith neurofibromatosis 1: consensus statement from the NF1 optic pathwayglioma task force. Ann Neurol 1997; 41: 143–149.

6 Bajenaru ML, Hernandez MR, Perry A, Zhu Y, Parada LF, Garbow JR et al. Opticnerve glioma in mice requires astrocyte Nf1 gene inactivation and Nf1 brainheterozygosity. Cancer Res 2003; 63: 8573–8577.

7 Zhu Y, Harada T, Liu L, Lush ME, Guignard F, Harada C et al. Inactivation of NF1 inCNS causes increased glial progenitor proliferation and optic glioma formation.Development 2005; 132: 5577–5588.

8 Bajenaru ML, Zhu Y, Hedrick NM, Donahoe J, Parada LF, Gutmann DH. Astrocyte-specific inactivation of the neurofibromatosis 1 gene (NF1) is insufficient forastrocytoma formation. Mol Cell Biol 2002; 22: 5100–5113.

9 Dahiya S, Lee da Y, Gutmann DH. Comparative characterization of the human andmouse third ventricle germinal zones. J Neuropathol Exp Neurol 2011; 70:622–633.

10 Doetsch F, Garcia-Verdugo JM, Alvarez-Buylla A. Regeneration of a germinal layerin the adult mammalian brain. Proc Natl Acad Sci USA 1999; 96: 11619–11624.

11 Yeh TH, Lee da Y, Gianino SM, Gutmann DH. Microarray analyses reveal regionalastrocyte heterogeneity with implications for neurofibromatosis type 1 (NF1)-regulated glial proliferation. Glia 2009; 57: 1239–1249.

12 Assanah M, Lochhead R, Ogden A, Bruce J, Goldman J, Canoll P. Glial progenitorsin adult white matter are driven to form malignant gliomas by platelet-derivedgrowth factor-expressing retroviruses. J Neurosci 2006; 26: 6781–6790.

13 Masui K, Suzuki SO, Torisu R, Goldman JE, Canoll P, Iwaki T. Glial progenitors in thebrainstem give rise to malignant gliomas by platelet-derived growth factor sti-mulation. Glia 2010; 58: 1050–1065.

14 Neymeyer V, Tephly TR, Miller MW. Folate and 10-formyltetrahydrofolate dehy-drogenase (FDH) expression in the central nervous system of the mature rat.Brain Res 1997; 766: 195–204.

15 Ozerdem U, Grako KA, Dahlin-Huppe K, Monosov E, Stallcup WB. NG2 pro-teoglycan is expressed exclusively by mural cells during vascular morphogenesis.Dev Dyn 2001; 222: 218–227.

16 Levine JM, Reynolds R, Fawcett JW. The oligodendrocyte precursor cell in healthand disease. Trends Neurosci 2001; 24: 39–47.

17 Zhu X, Bergles DE, Nishiyama A. NG2 cells generate both oligodendrocytes andgray matter astrocytes. Development 2008; 135: 145–157.

18 Lee da Y, Gianino SM, Gutmann DH. Innate neural stem cell heterogeneitydetermines the patterning of glioma formation in children. Cancer Cell 2012; 22:131–138.

19 Dawson MR, Polito A, Levine JM, Reynolds R. NG2-expressing glial progenitorcells: an abundant and widespread population of cycling cells in the adult ratCNS. Mol Cell Neurosci 2003; 24: 476–488.

20 Raff MC, Abney ER, Cohen J, Lindsay R, Noble M. Two types of astrocytes incultures of developing rat white matter: differences in morphology, surfacegangliosides, and growth characteristics. J Neurosci 1983; 3: 1289–1300.

21 Raff MC, Abney ER, Miller RH. Two glial cell lineages diverge prenatally in rat opticnerve. Dev Biol 1984; 106: 53–60.

22 Miller RH, David S, Patel R, Abney ER, Raff MC. A quantitative immunohisto-chemical study of macroglial cell development in the rat optic nerve: in vivoevidence for two distinct astrocyte lineages. Dev Biol 1985; 111: 35–41.

23 Stallcup WB, Beasley L. Bipotential glial precursor cells of the optic nerve expressthe NG2 proteoglycan. J Neurosci 1987; 7: 2737–2744.

24 Baracskay KL, Kidd GJ, Miller RH, Trapp BD. NG2-positive cells generate A2B5-positive oligodendrocyte precursor cells. Glia 2007; 55: 1001–1010.

25 Yoo S, Wrathall JR. Mixed primary culture and clonal analysis provide evidencethat NG2 proteoglycan-expressing cells after spinal cord injury are glial pro-genitors. Dev Neurobiol 2007; 67: 860–874.

26 Ju PJ, Liu R, Yang HJ, Xia YY, Feng ZW. Clonal analysis for elucidating the lineagepotential of embryonic NG2(þ ) cells. Cytotherapy 2012; 14: 608–620.

27 Mokry J, Karbanova J, Filip S, Cizkova D, Pazour J, English D. Phenotypic andmorphological characterization of in vitro oligodendrogliogenesis. Stem Cells Dev2008; 17: 333–341.

28 Winkler EA, Bell RD, Zlokovic BV. Central nervous system pericytes in health anddisease. Nat Neurosci 2011; 14: 1398–1405.

29 Wiranowska M, Ladd S, Smith SR, Gottschall PE. CD44 adhesion molecule andneuro-glial proteoglycan NG2 as invasive markers of glioma. Brain Cell Biol 2006;35: 159–172.

30 Al-Mayhani MT, Grenfell R, Narita M, Piccirillo S, Kenney-Herbert E, Fawcett JWet al. NG2 expression in glioblastoma identifies an actively proliferating popula-tion with an aggressive molecular signature. Neuro Oncol 2011; 13: 830–845.

31 Schrappe M, Klier FG, Spiro RC, Waltz TA, Reisfeld RA, Gladson CL. Correlation ofchondroitin sulfate proteoglycan expression on proliferating brain capillaryendothelial cells with the malignant phenotype of astroglial cells. Cancer Res1991; 51: 4986–4993.

32 Ellis JA, Castelli M, Bruce JN, Canoll P, Ogden AT. Retroviral delivery of platelet-derived growth factor to spinal cord progenitor cells drives the formation ofintramedullary gliomas. Neurosurgery 2012; 70: 198–204.

33 Wang J, Svendsen A, Kmiecik J, Immervoll H, Skaftnesmo KO, Planaguma J et al.Targeting the NG2/CSPG4 proteoglycan retards tumour growth and angiogenesisin preclinical models of GBM and melanoma. PLoS One 2011; 6: e23062.

34 Ogden AT, Waziri AE, Lochhead RA, Fusco D, Lopez K, Ellis JA et al. Identification ofA2B5þ CD133- tumor-initiating cells in adult human gliomas. Neurosurgery 2008;62: 505–514.

35 Tchoghandjian A, Baeza N, Colin C, Cayre M, Metellus P, Beclin C et al. A2B5 cellsfrom human glioblastoma have cancer stem cell properties. Brain Pathol 2010; 20:211–221.

36 Liu C, Sage JC, Miller MR, Verhaak RG, Hippenmeyer S, Vogel H et al. Mosaicanalysis with double markers reveals tumor cell of origin in glioma. Cell 2011; 146:209–221.

37 Alcantara Llaguno S, Chen J, Kwon CH, Jackson EL, Li Y, Burns DK et al. Malignantastrocytomas originate from neural stem/progenitor cells in a somatic tumorsuppressor mouse model. Cancer Cell 2009; 15: 45–56.

38 Ono K, Yasui Y, Rutishauser U, Miller RH. Focal ventricular origin and migration ofoligodendrocyte precursors into the chick optic nerve. Neuron 1997; 19: 283–292.

39 Gao L, Miller RH. Specification of optic nerve oligodendrocyte precursors byretinal ganglion cell axons. J Neurosci 2006; 26: 7619–7628.

40 Tchoghandjian A, Fernandez C, Colin C, El Ayachi I, Voutsinos-Porche B, Fina F etal. Pilocytic astrocytoma of the optic pathway: a tumour deriving from radial gliacells with a specific gene signature. Brain 2009; 132(Part 6): 1523–1535.

41 Silber J, Lim DA, Petritsch C, Persson AI, Maunakea AK, Yu M et al. miR-124 andmiR-137 inhibit proliferation of glioblastoma multiforme cells and induce differ-entiation of brain tumor stem cells. BMC Med 2008; 6: 14.

42 Kwon CH, Zhao D, Chen J, Alcantara S, Li Y, Burns DK et al. Pten haploinsufficiencyaccelerates formation of high-grade astrocytomas. Cancer Res 2008; 68:3286–3294.

43 Abel TW, Clark C, Bierie B, Chytil A, Aakre M, Gorska A et al. GFAP-Cre-mediatedactivation of oncogenic K-ras results in expansion of the subventricular zone andinfiltrating glioma. Mol Cancer Res 2009; 7: 645–653.

44 Zhu Y, Romero MI, Ghosh P, Ye Z, Charnay P, Rushing EJ et al. Ablation of NF1function in neurons induces abnormal development of cerebral cortex andreactive gliosis in the brain. Genes Dev 2001; 15: 859–876.

45 Madisen L, Zwingman TA, Sunkin SM, Oh SW, Zariwala HA, Gu H et al. A robustand high-throughput Cre reporting and characterization system for the wholemouse brain. Nat Neurosci 2010; 13: 133–140.

46 Brannan CI, Perkins AS, Vogel KS, Ratner N, Nordlund ML, Reid SW et al. Targeteddisruption of the neurofibromatosis type-1 gene leads to developmentalabnormalities in heart and various neural crest-derived tissues. Genes Dev 1994; 8:1019–1029.

47 Hegedus B, Banerjee D, Yeh TH, Rothermich S, Perry A, Rubin JB et al. Preclinicalcancer therapy in a mouse model of neurofibromatosis-1 optic glioma. Cancer Res2008; 68: 1520–1528.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)


11


http://www.nature.com/onc

ng2-cells are not the cell of origin for murine...

Documents